Interdisciplinary Center Psychopathology and Emotion Regulation
University Medical Center Groningen
University of Groningen/University of Amsterdam
Advancing Personalized Network Modelling in Clinical Practice
The field of network psychometrics has gained increasing importance over the last decade (Robinaugh, Hoekstra, Toner, & Borsboom, 2020; Cramer et al., 2016; Borsboom, 2017). Further, statistical tools can now be applied to individuals: In psychotherapy, personalized network modeling can have great potential to inform many decisions (e.g., case formulation and treatment selection; Epskamp et al., 2018). However, the current practice of estimating personalized network models from clinical data based on vector-autoregressive models face several limitations: The number of observations usually obtained in clinical practice can lead to underpowered estimation of the type of models, while relevant knowledge is available on the side of the clinicians. Further, it is unclear how to precisely answer clinically relevant questions given clinical data, most importantly in respect to the selection of treatment targets. A recent study proposed a methodology based on control theory (Henry, Robinaugh, & Fried, 2020), such methods will be an important avenue to advance personalized network models in clinical practice.
The project is granted by the NWO research talent scheme and focusses on tackling these issues by proposing new methodologies. The planned outcomes are: (1) Developing a prior elicitation module for clinicians to specify a prior networks, with consequently providing a method to update prior beliefs with ESM data using Bayesian inference. This is addressing power issues (less observations needed, as well as implementation.) (2) Advancing computational modeling in clinical practice using differential equation modeling, which could help in informing the design of ESM studies, such as which variables to select, or which time-intervals should be modeled. During the first phase of my PhD, I wrote a paper on this issue that is in press at BMC Medicine (Burger et al., 2020). (3) Implementation of elements in control theory for personalized network models that can give an indication for treatment selection.
Prof. dr. R.A. Schoevers, dr. S. Epskamp, dr. C. van der Veen
NWO Research Talent Grant